from datasets import load_dataset

data = load_dataset("csv", data_files={"train": "train.csv"})
data = data["train"].train_test_split(test_size=0.1)  # 90/10 split